Room equalization - FFT in concert

f ■:■:::-■.■ :-:V * ... * I Touring bands present their | engineer with what often seems like an insurmountable .v""'""""""-:^ ■.....::::■'■■■■■ ■ ■■■■■ ■ ■"■J ^\4 *'*r< 1 LLL 1.1 problem - how to produce good quality sound in a multitude of different and varying acoustic environments. The engineer's experience alone will not always produce the best results, if the tools he 1 t L L L 444 4 " " " T 4 4 4 4 T4 i i i . 4444 4444444444 "111 444 44 4 wwwwwwwww .,. . . . ., . . . .,, ,^. . .„ ,^- .//, ,*,- 4444^ -444444 444444 444444 444444 44444 444 ^^ LL.'_4 444 4**4 444 4* 44444 444444 4444' *T ' ' ' '* * 4T ' ' '# *^T ' '# + + 4 " '^# #*> ■■■■■■■■■■■■■■■■■■■■■■ ■■■■■■■■■■ »«%» •■ # #«*« 4* »■ * *4 ' '^4 *'*_•' ' '4 ■" " "J " " " " ' ^" ' f * ' ' ' ■ ' S S .-.- ' ' S . ^ 4" ' S . ■_^ ' rf- ■ ■ ■ " '

has to use are not flexible - L . ■ ■ L L L . ■ L L L . ■ 4 4 4 4 ' ' ' L ' L ' L L L ' ' ' ■ L L ' ' ' ' ■ L L .' ' ' ' .' ' ' ' ' ' ' ' ' ' ' ' ' ' ' '' ' ' ' ' ' ' '' ' ' ' ' ' ' '' ' ' ' ' ' '' ' ' enough to cope with every situal tion :::- ::::: ■.:::'.■■.::'.■■..■.:■...■:::■..:::■.:::: ,::\\: \\\\\■;■ ■ ■;".:"f:■V" :.V::".:.v."::»:i:..v■ |.:..;:::■:.::::.' |,::::■;■;■::::■;■=■=■:::■-■=■::::_-:^?:-:_:■ :-:-:->::■:/--:-^:-f:■:■-■":::::""■":::V"f::V_:_:?::"::-:-:?:":^<:-:f::<:-:^::"::--L:^-:f: An FFT analyzer has theIverlll T TT T T ■ 444 444444 4 4 4 4 444 ''444 ''4444 4*4 4 4^ ^ L 4 4 4 ■*»'J ***** 44444 44444 4444 44 »""»"" »»»»» — »»»»» ^^T . .,, ^^T ..,,,... .,, ^^T f| 4 ,\'/, 444 ,'i 4 satility needed, not only to enJl! sure that the engineer s skill is

...... x ■■■■:■■■ x ::■■::: .-.■■■■..■.■■■•...■.■■....■.■• ■ ■...■.■■.■...■.■.■...■.■■...■,„.■. + ■■.-.■.— ■■.■. ■. ^^ ■. .-.■■■■ ^ .-.-.■■,...-.- ^....-.- •...... ■ II not wasted, but also to help in L LL' ■ L L L' ■ ■ L L LL' . L L L L' ■ L L ' ' ' ' ^' ' ' ''+^ ' ■■■...■■■ / . . '. " " / ^ L L ' ' L L L ' ' ■ ■ L L ' ' ' ■ L L L ' ' ' ■ L ' ' ' '. ' ' ' ^ . . '. '. ' ' ^ . . . '. ' ' ^ . . . '. ' ^ . . . . '. producing better sound quality. \ / V" * *+ "*:* * *++ ■*■:?■■"::*■■*:::*■■■::::: ■,://.■ -,X T^; ;

The problem. In general speaker-room response is 2034) is a dual-channel Fast Fourier measured using a 1/3-octave analyzer, Transform (FFT) analyzer. It is de- Live sound can be dead. The sound and the observed resonance peaks and signed to investigate cause and effect are then equalized with 1/3- relationships, in any system where justed and equalized before the con- or 1/6-octave graphic-equalizers. Un- both the cause and effect are electrical cert, is now boomy, dull and hollow. fortunately, many low-frequency rum- signals, or can be converted to electri- The band is discouraged, the audience ble and mid-range ringing problems cal signals by transducers such as mi- disappointed, and the sound engineer can be caused by narrow-band reso- crophones. The 2032 samples, ana- suffers sheer frustration for the next nances which are too narrow to be lyzes and averages the data in both few hours. So what went wrong? measured in a 1/3-octave analysis, nor signal inputs and, by comparing the can they be corrected by fixed band- two frequency spectra, computes the The problem is that a live concert is width graphic-equalizers. frequency response and other charac- very dynamic and the acoustics of the teristics of the system. Because the venue can change dramatically. Con- The use of live music as the test 2032 operates on both signals, and ventional techniques for adjusting signal, adjustable equalizers, and an uses the system input as the reference equalization (EQ) use clicks, chirps or analyzer able to pin-point narrow- for analysis the technique is indepen- pink-noise as the test signal; these are band resonances would be the ideal dent of the test signal. Therefore any obtrusive sounds and cannot be used solution. This would mean that the test signal, including music, can be once the audience is present. The EQ could be adjusted for changes in used. room, however, becomes an acoustical- the room acoustics when the audience ly different place when filled with peo- arrives, and corrected for further pie - sound distribution, reverbera- changes during the performance. Usill£ 3.11 FFT tion-times ■■ and floor-reflections change. The humidity and tempera- An FFT analyzer, used in conjunc- The 2032 is easily connected to ture variations caused by the audience tion with high quality parametric- sound-reinforcement systems to mea- and the stage lighting will alter the equalizers, provides the answer. sure room-response (Fig. 1). For the sound-absorption characteristics of measurements you need two signals, the air. The net result of all these are The answer. the system input for channel A of the shifts in room resonances which can * analyzer is taken from the output of ruin the EQ. The Bruel&Kjasr Type 2032 (or the mixing console. The channel B in-

BO 0167-11 2C 50 'CO 200 500 *k 862109 Fig. 4. Figs. 2 & 3 combined to show the fit between the room-response and the equalizer

Fig. 5. Room-response measurement with equalization

Fig. 1. A block diagram of the dual-channel analyzer connected to the sound system. Measureing between the A and B input signals, with music as the test signal, the ±YL pGriOrillcinCG. analyzer computes the room-response Whilst the audience is arriving, you can check the EQ using recorded mu- put is taken from one of two sources in nal delay line, and can also be used to sic as the test signal. Just before the turn, firstly the output from a cali- time-align the speaker system. concert starts, therefore, the EQ will brated microphone located in the au- have been adjusted for a full house. dience, usually in the house-mix posi- A frequency response measurement The FFT enables you to continue tion, and secondly the output of the without EQ will then give you the monitoring the speaker-room response equalizer. speaker-room response (Fig. 2). This during the performance, using the live measurement can be stored and dis- music for the measurements. Continu- An initial frequency response mea- played whilst, with the delay switched ous corrections can be made to the EQ surement has to be made, between the off, the inverse frequency response of as acoustic conditions change. console output and the calibrated mi- the equalizer is continuously mea- crophone, with live or recorded music sured and displayed in real-time on Audience reaction and other ran- as the test signal. From this measure- the split-screen. This enables you to dom noises will tend to spoil the mea- ment you can calculate the time delay set up the EQ (Figs. 3 & 4) to compen- surements, but here the coherence between the speaker-system and the sate for the speaker-room response, function on the analyzer will help you. microphone. This enables you to set and to produce the overall response Coherence is a measure of how well the time delay into the analyzer so you want. A further speaker-room re- the data in channel B is correlated that the channel A data is delayed, sponse measurement, using the micro- with that in channel A. Looking at the until the microphone response reaches phone and time delay, with EQ ap- coherence, whilst measurements are channel B. The built-in user-defined plied will then show the effect your being made, you can see data averag- delay eliminates the need for an exter- equalization has had (Fig. 5). ing reducing the effect of random noise. Continue the measurements un­ til you are sure that enough averages have been taken for good coherence, and therefore a valid response mea­ surement.

Great sound gets great audience re­ action - 2032 works with it and helps you get it.

Fig. 2. The room-response measurement Fig. 3. The inverse frequency-response The measurements shown were made in the without equalization measurement of the equalizer World Theater.